documentation
01 get started

Installation

Exolvra runs as a single self-contained application. Once it's running, the whole product lives in your browser — the dashboard opens on port 5079, and there's nothing else to stand up.

Before you start

This page covers first launch and verifying the install. How you obtain the Exolvra package depends on your setup — many people are handed a ready-to-use instance by whoever administers it, in which case you can skip straight to First login with the address and bootstrap password they gave you.

You need very little to run Exolvra:

  • An LLM provider API key — Anthropic, OpenAI, Google Gemini, Groq, Mistral, and others are supported. You add this from the dashboard after first launch, so you don’t need it in hand yet. If you’d rather not pay anyone, a local Ollama install gives you a free option.
  • Nothing else is required. Exolvra does not need Docker, Redis, or PostgreSQL. The default storage is a single SQLite database under your user home directory.

Two things are optional and only matter if you plan to install community MCP servers:

  • Node.js LTS — for MCP servers that ship as Node packages. See MCP setup.
  • Python 3.10+ — for Python-based MCP servers.

First launch

Start Exolvra. On its first run it creates its database, sets up a default admin account, and prints a one-time bootstrap password to its console output. Copy that password — you’ll need it to sign in, and it isn’t shown again.

When Exolvra is up it serves the dashboard locally:

Now listening on: http://localhost:5079

Open http://localhost:5079/ in your browser. You’ll land on the onboarding wizard — First login walks through it.

Where your data lives

Exolvra keeps everything under one folder: ~/.exolvra/ on Linux and macOS, or %USERPROFILE%\.exolvra\ on Windows. That single folder holds the SQLite database, agent workspaces, generated files, and encrypted credentials.

Back up that folder and you’ve backed up your entire instance. To move an installation to another machine, copy the folder across and start Exolvra there — it picks up exactly where it left off.

If you want the data somewhere else, set the EXOLVRA_DATA_DIR environment variable to a path of your choosing before starting Exolvra.

Verifying the install

Once the dashboard is up, three quick checks confirm everything is wired:

  1. Visit http://localhost:5079/ → you should see the Exolvra landing page or the onboarding wizard.
  2. Visit http://localhost:5079/health → returns JSON with readiness status for the database, budget, and providers.
  3. Sign in with the bootstrap credentials, go to Settings → Providers, and paste an API key for one provider (Anthropic, OpenAI, etc.). Save. Open Chat, pick the default agent, and send a message — if you get a reply, your install is live.

Keeping it running

For anything beyond casual use you’ll want Exolvra to restart automatically after a reboot. The standard approach is to run it under your operating system’s service manager — a systemd service on Linux, or a Windows Service — with its data directory (~/.exolvra, or whatever you set EXOLVRA_DATA_DIR to) on persistent storage. Point the service manager at the Exolvra application and let it handle restarts.

Next

  • First login — walk through the setup wizard and change your admin password
  • Your first agent — create a specialist from the dashboard and give it work
  • Admin overview — the five-minute admin checklist to harden a new instance